I'm assuming decent initial placement (EVERYONE gets a pick at one of the 4-best spots in the game, and the player who get's 4th pick gets 4th AND 5th pick). Everyone has at least one good pick in my experience.

The situation where you have a clear "dead man walking", even when you did everything correctly, is when opposing players block you in with roads, preventing your expansion.

Once one of your initial 2 villages are blocked, the 2nd village becomes a target for blockage, because everyone knows that blocking your 2nd village will effectively eliminate you from the game.

Not "really eliminate", you're still a player and are kinda forced to play. But you can't make a 3rd village, and basically have no pathway to victory. A few unlucky rolls, or an opponent's lucky "Road Building" card, can bring up the road-block strategy extremely quickly.

Catan can get pretty uninteresting due to dice variance. I've had games where two players made lots of resources on 9 and two players made lots of resources on 5. Five got rolled 14 times and 9 got rolled 3 times. Even with good spots if the dice are against you it can easily become a lot of do nothing while other people play the game.

A lot of people also play with the houserule no trading because there is too much kingmaking in trading. It's too easy to place aggressively for ore if you know you can trade for whatever you're missing of the settlement. If you have to 4:1 it's a bit more of a tradeoff.

Wait, no trading??? That seems to remove the only interesting aspect of the game!

Otherwise you're all sitting around a board that randomly generates resources for you to spend and sorta get in each other's way with road and settlement placements. If you want semi-solo Euro games, play something with way less dice variance!

The trades and dynamic that arises from the mechanic are pretty much the only reason to play imo. Otherwise the dice mechanics are way too strong as you point out.

When we played Catan every single day for a year during the 1990s, we ran into the dice variance issue plenty of time that we considered creating a bag with 36 chits with the dice distribution, guaranteeing that all dice results would come up in the expected ratios. Or maybe 72 chits to remove the predictability when the bag starts to get empty, and return the chits once you got a complete set of 36 out.

But as you can see, the idea quickly became way too complicated and we never implemented it.

Removing trading sounds like a terrible idea. Being able to monopolise a resource just to trade it for what you really need, is a feature, not a bug. Some resources are rare, and you need to take that into account during placement.